On hockey, analysis, and narrative

One of the first things I learned in graduate school was that history isn’t really a collection of facts about the past, although that’s the way it’s taught in American schools. History is really a method for finding new & better questions to ask about the past. Because the question you ask has a profound impact on the picture that emerges. Modern American historians have as their disciplinary bedrock an awareness of the explanatory power of narratives.

See, back in the 1970s, Hayden White and others started talking about narratives and their power to change what we define as reality. Human experience, these historians said, was far too complex to ever be fully described. Trying to capture everything results in chaos instead of understanding. In order to make sense, we had to leave things out. The process of choosing what to leave out changed the tone and thrust of the final narrative.

In other words, narratives by their very nature are explanations of the world. They are replete with cause and effect, with value judgments, with implications about what and who is important to the story and why. Any conversation that includes narratives is limited by the parameters that those narratives–consciously and unconsciously–provide.

At the same time, narratives are deeply rooted in the structures of the human brain. Narratives are one of the first things proto-humans did with the language they were developing. They told weather stories and hunting stories and travel stories and heavens stories. They told stories about the afterlife. They told stories about cause and effect and power and victory and good and bad. Narrative makes the world livable, but it does so by enforcing an artificial organization upon our experiences. And that organization can be changed when one fixes upon a different set of details.

Think, for instance, of a job search. When you’ve gone on several job interviews without getting a job offer, it’s natural to try to understand why. Is it your credentials? Your resume? Your body language? Your age, race, or sex? Your job history? Your personal hygiene? Which of the hundreds of thousands of details are the most salient to your failure to get a job? Which are irrelevant? How can you change your strategy to finally, finally get that offer? A new suit? A power tie? A good luck charm? What?

It is most likely that there isn’t a single thing that has had an outsized effect. The most likely answer is that for each interview a different detail mattered most. That is supremely unhelpful, however, as it provides you with no plan of action to maximize your chances the next time. And it makes you feel completely out of control as well.

So the search for an answer, a single, helpful answer will go on, even if you don’t want it to. You dwell on the situation until you have some kind of plan that makes you feel less unmoored and at the mercy of remote and impersonal forces. It may not be the right answer. It may not even be a good answer. But it’s better than the idea that there is no answer. Narrative is explanation and explanation is control. But it requires leaving out information, some of which probably ought not to be left out.

This understanding of narrative–that it is explanation, that it is by its very nature incomplete, and that it is easily manipulable and at times profoundly self-contradictory–is at the core of the modern discipline of history. We’ve been grappling with the question of narratives for generations now and have to some extent made peace with the idea that we can’t discard them, but we do need to be careful about building them.

I’d posit that this awareness of the power and flaws of narratives also underlies the project of hockey analysts who embrace the newer “advanced” statistics like shot-based possession metrics, PDO, and zone entries. While those who are most prominantly and vocally against them see narratives as wholly unproblematic.

The “anti-stats” folks are more likely to see their narrative as natural and uncontrived, a self-evident, complete, and unbiased picture of the experience of hockey. They throw around cause and effect like candy. They assume that the connections between the details they choose to leave in and the end result of a game or a season are not only logical but ineluctable. They are so natural as to be unassailable and inescapable. Because the narrative is natural, there’s no need for details that just confuse the situation.

Sometimes, the explanations are plausible. Certainly they have for decades given people who make decisions about hockey plenty of guidance. And sometimes that has paid off.

Oftentimes, though, it hasn’t. The narratives often don’t match up to external records, but equally as important, they are internally weak as well. For every event, multiple causes–sometimes contradictory–can be given, depending on which causal connection a writer wants to emphasize. At the same time, even within the world of a narrative, a single cause can have multiple and sometimes mutually exclusive effects. Over time, the traditional narratives undermined the very foundation of their authority to determine the explanations of hockey, just as they did for history those many decades ago.

The stats project, on the other hand, comes from an understanding of the flaws in narrative–flaws that reside not in the act of choosing details but in the naturalization of the choices made. People who use these kinds of stats to build their narratives appear to have a stronger awareness of the effect of choosing on our ability to comprehend our experience. They want, they need, to ask better questions so that they can get answers that give better plans of action.

The stats project started as an attempt to discover which details matter, independently of traditional narratives. To consider, test, and weigh the importance of plays that were not assumed to be self-evidently effective. And they latched onto the idea that narrative, like bureaucracy, has a life of its own. That word gets used often in hockey circles to mean “bias” or “unworkable explanations.” In actuality, stats folks are not escaping narratives but are rewriting them with an emphasis on different details.

It is hoped that over time, the new narratives of hockey will bring better answers. Even more, I think, it is hoped that by challenging old narratives, hockey analysis can retain curiosity about cause and effect and remain open to the awareness that no explanation is perfect or natural. We should continue to keep pushing, keep seeking, keep asking questions. In other words, let the new hockey analysis become a way of discovering new and better questions to ask rather than a repository of facts about hockey games.